Hi,

I am performing DiffBind on ATACseq data and I have a question regarding the use of block and design in setting a dba.contrast. From a previous question a couple of years ago I learned how to use block (DiffBind paired samples) but I think it might be different now in the new version as the design setting has changed.

I would like to block for replicate (or individual patient in my case) whilst also using a design that includes Tissue as confounder: ~Tissue + Condition. However, when I do that it says that block is ignored when design is used. Should I just include replicate in my design instead of as block or is there another way to analyze my data in a paired way while also including confounders in the design?

Data <- dba.contrast(Data, block = DBA_REPLICATE, design = "~Tissue + Condition", reorderMeta = list(Condition = "Isolated")) 

Output:

Computing results names...
Warning message:
block ignored when design is used.

Many thanks for any help!

The original block parameter was there to get around DiffBind not supporting arbitrary design formulae. It has been retained for backward compatibility.

The best way to do this now is to put everything (including Replicate) explicitly in the design. So don't use block, and set design="~Replicate + Tissue + Condition".